From owner-freebsd-questions Fri Apr 24 02:13:03 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id CAA00332 for freebsd-questions-outgoing; Fri, 24 Apr 1998 02:13:03 -0700 (PDT) (envelope-from owner-freebsd-questions@FreeBSD.ORG) Received: from freebie.lemis.com (freebie.lemis.com [139.130.136.133]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id CAA00326 for ; Fri, 24 Apr 1998 02:12:57 -0700 (PDT) (envelope-from grog@lemis.com) Received: from papillon.lemis.com ([192.122.138.250]) by freebie.lemis.com (8.8.8/8.8.7) with ESMTP id SAA00859; Fri, 24 Apr 1998 18:42:14 +0930 (CST) (envelope-from grog@lemis.com) Received: (grog@localhost) by papillon.lemis.com (8.8.8/8.6.12) id KAA00595; Fri, 24 Apr 1998 10:17:16 +0800 (SGT) Message-ID: <19980424101716.53779@papillon.lemis.com> Date: Fri, 24 Apr 1998 10:17:16 +0800 From: Greg Lehey To: Vallo Kallaste , freebsd-questions@FreeBSD.ORG Subject: Re: ccd performance limit? References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.89i In-Reply-To: ; from Vallo Kallaste on Wed, Apr 22, 1998 at 08:08:16PM +0300 WWW-Home-Page: http://www.lemis.com/~grog Organization: LEMIS, PO Box 460, Echunga SA 5153, Australia Phone: +61-8-8388-8286 Fax: +61-8-8388-8725 Mobile: +61-41-739-7062 Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Wed, 22 April 1998 at 20:08:16 +0300, Vallo Kallaste wrote: > > On Wed, 22 Apr 1998, Hans Huebner wrote: > >> When I set up a ccd device with the four disks, I am unable to read more >> than 14-15 MB per second from the stripe set. I tried various different >> interleave factors (and found that 36 seems to be the optimum value). > > I'm also experimenting with ccd at this time and also get only 14-15 MB/s. > I have two Quantum Viking disks and tried different interleave factors, > too. Looks like almost same problem :( The obvious question is: how are you using the stripe set? How are you measuring the throughput? ccd has a certain overhead, of course, but part of its performance increase is due to the ability to server multiple requests in parallel. If you measure only a single requester, you will get results which may not have much significance in practice. If you are measuring multiple access, I'd be interested in knowing how you measure the throughput. I'm currently writing a replacement for ccd. If you're interested, you can test it. I'm not guaranteeing higher throughput, but I'd be interested in comparing it on high-performance disks (I'm developing on the oldest disks I can find, so that I can pinpoint performance problems). Greg To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message